WORD

ISSN: 0043-7956 (Print) 2373-5112 (Online) Journal homepage: https://www.tandfonline.com/loi/rwrd20

Etymology and General Linguistics

Yakov Malkiel

To cite this article: Yakov Malkiel (1962) Etymology and General Linguistics, WORD, 18:1-3,
198-219, DOI: 10.1080/00437956.1962.11659774

To link to this article: https://doi.org/10.1080/00437956.1962.11659774

Published online: 04 Dec 2015.

Submit your article to this journal

Article views: 2343

View related articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=rwrd20
YAKOVMALKIEL-----------------------

Etymology and General Linguistics

1. The anomalous status of etymology. Within the bundle of linguistic
disciplines etymology occupies a position difficult to define. It does not,
strictly speaking, mark a transition between the domains of language
viewed chiefly as a means of communication (linguistics) and language
considered as a vehicle of art (lite;ature), in the way poetics, stylistics, and
folklore may be interpreted; neither does it bear any resemblance to se·
mantics. Of every other discipline (say, phonology and syntax) one can
imagine a synchronic and a diachronic projection. But synchronic etyrno·
logy-despite the authority of an aging Vendryes (Bulletin de Ia Societe de
linguistique de Paris XLIX: 1 [1953], 1-19: "Pour une etymologie statique")
-amounts to scarcely more than a paradox. For approximately a century
scholars have been operating with the term and concept of folk etymology,
which again lacks any counterpart in other disciplines.

2. Conflicting views. An appeal to leading authorities to clarify the situa-

tion is of little avail, since their verdicts have been either evasive or con·
tradictory. In some slimmer introductory treatises, from the eleven loosely
strung essays forming Sapir's Language (1921) to the six stimulating, if
uneven, chapters which Martinet has combined into his Elements de
linguistique generale (1960), etymology barely, if at all, receives incidental
mention. The situation is somewhat different, but hardly less disappointing,
with the broader outlines. Even that most lucid of theorists, Saussure, who
relegates-surely not by chance-his brief discussion to an Appendix, pre·
sents a picture not wholly convincing. For him etymology is neither an
autonomous discipline nor a smoothly integrated part of evolutionary
linguistics: it amounts to a special application, to early stages of word
history, of principles generally valid in linguistic research. Saussure's prime
examples are four pairs of words-all perfectly transparent-illustrating,
in this order, sound change, semantic change, simultaneous change of
sound and meaning, and derivation. Later, more complex relationships
come up for mention, e.g., the link of Fr. oiseau to Lat. aui(cellu)s; the
(historical) study of suffixes and prefixes is likewise included. After a some-
what lukewarm remark to the effect that explaining one word means tracing
198
 ETYMOLOGY AND GENERAL LINGUISTICS 199

it to other words, Saussure recognizes certain ties between etymology, on

the one hand, and such better established disciplines as phonology, mor-
phology, semantics, on the other, but doubts that any strict methodology
can be prescribed and expressly discards the possibility of rigorous opera-
tional arrangement. 1
Bloomfield's approach (1933) is equally unpromising. His book contains
no single section, let alone chapter, on etymology pure and simple, but the
Index leads one to some relevant passages, as when etymology is identified
as a special concern of ancient Greek scholars (4) and, later, as a butt of
Voltaire's sarcasm (6). More important is the modern "precise" definition
of the etymology (15) of a speech-form as "simply its history ... obtained
by finding the older forms in the same language and the forms in related
languages which are divergent variants of the same parent form" (15); the
tracing of Engl. mother via OE modor and its congeners to Prim. Gmc.
*moder, hence to Prim. IE *miiter serves as a prime example. In accordance
with this definition, the technique of etymology is obliquely mentioned
first under Phonetic Change (346-347, 351-355), later under Semantic
Change (427-430).
The dual disadvantage of such a position consists in this: (a) etymology
is reduced, as it were, to a small-scale-indeed, the smallest-operation
within diachronic phonology, a role which deprives it (to make things
worse, inexplicitly) of any independent status; (b) the reader remains un-
aware of the fact that professional etymologists will rebel against examining
problems as transparent as the provenience of mother. Surely, it is no
mere coincidence that a handbook as influential in this country (and
later the world over) as Bloomfield's has singularly failed to stimulate
the slightest curiosity about genuine etymological research, at its most
exciting.
1 "Comme Ia linguistique statique et evolutive, elle decrit des faits, mais cette des-
cription n'est pas methodique, puisqu'elle ne se fait dans aucune direction deter-
mince .... Pour arriver a ses fins, elle se sert de tous les moyens que Ia linguistique met
asa disposition, mais elle n'arrete pas son attention sur Ia nature des operations qu'elle
est obligee de faire" (2d ed. [Paris, 1922], p. 260; Supplement C to Parts III-IV).
Between 1893 and 1912 Saussure taught several formal courses on etymology, with
emphasis on Greek and Latin, and his ideas on the scope of the discipline must have
wavered considerably; seeR. Godel, Les sources manuscrites du "Cours de linguistique
generate" (Geneva and Paris, 1957), pp. 24, 25, 134. In his earlier synthesis Le langage
(1921; completed in 1914), p. 206, Vendryes equated etymology with diachronic lexi-
cology; what he accomplished in 1953, under the avowed influence of Saussure and some
ancient Indian grammarians as interpreted by F. Edgerton, was to substitute 'static
etymology' for synchronic lexicology, a less than felicitous decision. Meillet's etymo-
logical testament (1932) will be found in his prefaces to the Dictionnaire etymologique de
Ia langue latine and to 0. Bloch's Dictionnaire etymologique de Ia langue franfaise.
7••
200 YAKOV MALKIEL

3. Four claims to autonomy. The correct place of etymology, if one

agrees to define it as the search for word origins, must be sought elsewhere.
If it is true that word biographies lend themselves to graphic projection,
then the etymologist's task is the elucidation of the starting point, better
still, of the initial segment of chosen lexical trajectories. Etymology is thus
a mere subdivision of lexicology (here taken to mean 'theoretical, pre-
eminently historical study of the lexicon', in contrast to lexicography,
viewed as an applied science 2), but a subdivision endowed with several
peculiarities which tend to give it special rank. One can single out at least
four such claims to autonomous high status:
(1) Though classifiable by present standards as a mere subdivision, ety-
mology boasts a venerable history.of its own, throughout Antiquity and the
Middle Ages, a history involving significant contacts with areas of human
endeavor unrelated to lexicology and even to linguistics, as that science is
currently understood. Compressed into a formula, the paradoxical situa-
tion may be described thus: the part is older than the whole.
(2) Because creative etymology presupposes, on the part of its practi-
tioner, a desire to transcend the domains of the obvious and of the highly
probable in the matter oflexical equations and to operate in the hazardous
realm of the increasingly conjectural, it often attracts a totally different
type of personality than does grammar, even in its modern garb of
'structure'.
(3) Like all lexical subdisciplines, etymology is equally concerned with
form and with meaning and, through the latter, also with the outer world
of realities. But in a more intimate sense than the others, this subdiscipline
is tied up with certain facets of historical grammar, chiefly phonology and
derivation-a point duly recognized, but magnified out of all proportion,
by Saussure and Bloomfield.
(4) The main idiosyncrasy of etymology stems from the fact that, unlike
most cognate subdisciplines, it operates consistently with fragmentary
2 Not all lexical monographs exhibit an etymological slant. H. and R. Kahane's article
on the (predominantly nautical) progeny of surgere (Romance Philology IV [1950-51],
195-215) embodies an experiment in spatio-temporal semantics. The prime purpose of
my reconstruction of the Hispanic branch of per-, re-, suc-cutere (Hispanic Review XIV
[1946), 104-159) was to demonstrate that Class. and Mod. acudir-semantically many-
faceted, hence genetically elusive-perpetuates OSp. recudir. In special instances, as
where a blend is involved, heavy documentation seems indispensable; cf. OSp. Gal.-Ptg.
desmazelado 'wretched' (from mac-ula, -ella 'spot') x Hebr. maz·al 'star, destiny'> (Jud.-)
Sp. desmazalado 'weak, destitute' (Hispanic Review XV [1947), 272-301); but consistent
use of massive illustration, as advocated and very effectively practiced by W. von
Wartburg (on his technique see Word X [1957), 288-305; also, in his Bibliography [1956),
items A 138, 209, 350; B 19, 21), tends to smother the nuclear problems of etymology.
 ETYMOLOGY AND GENERAL LINGUISTICS 201
evidence, with dotted evolutionary lines. Every etymologist protests that
he would prefer to rely, in his reconstructions, on a vastly increased stock
of recorded forms; few would be candid enough to admit that truly com-
plete records would deprive the etymologist's endeavors of their real charm,
even of their raison d' etre. The sparseness or even unavailability of critically
needed material has fascinated some workers (moulding, in the process,
their personalities) and has, with equal power, repelled others.

Were it not for these four considerations, particularly the last three,
etymology could be safely eliminated from the roster of legitimate lin-
guistic pursuits and stored away as a curious relic of prescientific concern
with language.

4. A science or an art? Some observers argue that the place of etymology

within the alliance oflinguistic disciplines has been jeopardized through the
linguists' growing endeavor to attain the status of exact scientists. The two
main obstacles confronting etymologists who strive to keep pace with
fellow linguists so oriented, 3 these critics contend, are, first, the element of
haphazardness inherent in their discoveries; second, the fact that their re-
searches and working habits suggest an art rather than a science.
Either objection has merit, but loses something of its weight if carefully
qualified. In apprising both, it is helpful to delimit, more sharply than has
been done in the past, three phases of etymological inquiry: (a) the pre-
paration or training; (b) the actual discovery; (c) the presentation of
results.

5. The 'accidental' ingredient. Undeniably, there is an ineradicable ele-

ment of casualness at the actual stage of discovery; it follows from the
invariably fragmentary character of the material available. The etymologist
~ike every other archeologist) may litefally stumble upon the missing link:
an intermediate form deemed hypothetical may appear in a text recently
discovered or freshly re-read, a trace of the uncontaminated product
justifying a hazardous conjecture emerges in a newly surveyed dialect, etc.
Possibly more arresting and conducive to the injection of some such term
as 'intuition' are etymological identifications involving an abrupt semantic
change or a bold metaphor. These are sometimes made through direct con-
tact with nature (in a museum or in the open air), through increased
3 For the sake of the argument, etymologists are portrayed as forming, so to speak, a
separate task force. In reality, the same scholar often acts now as etymologist, now as
gmmmarian, now as textual critic; the greater his elasticity-if balanced by seriousness
of purpose-the more significant and the less lopsided his total contribution.
202 YAKOV MALKIEL

familiarity with the trades and with rural living, through immersion in past
ideas, beliefs, and sentiments, or through felicitous associations with
neighboring cultural climates and linguistic areas.
But Spitzer's familiar bon mot: "Suche keine Etymologien; finde sie!"
must not be taken at its face value. Though the actual 'Einfall' may be an
instantaneous, unforeseeable event (does not a similar situation-a sudden
flash of imagination-prevail in the physical sciences?) and though in ety-
mology certain mental qualities associated with creativeness, memory,
vividness of association, and even visual impressionability play a part at
least as crucial as that of straight indoctrination (and conceivably more
appealing to the sensitive layman), it is nonetheless true that important
phases of etymological inquiry may and should be placed under rational
control. Significant conjectures are"not known to occur to the uninitiated ;
it takes a mind not only plastic and versatile, but thoroughly attuned to
pending etymological problems (as a rule, through long, systematic ex-
posure to specialized teaching or to technical literature) to respond at once
to the challenge of a 'hunch'. No less important is the slow, predominantly
rational filtering of one's instantaneous insights, and in such final decisions
as whether to publish the new solution as a separate venture or to make it
part of an intricate strategy of long-term research, analytical thinking be-
comes the determining factor. Of the three phases of inquiry established in
the preceding Section, only Phase (b) shows a strong, apparently irreducible
admixture of the accidental.

6. The artistic element. Far more elusive, hence difficult to circumscribe,

is the artistic constituent of etymology, partly referring to the conduct of
the inquiry and to the comportment of its practitioner, partly crystallizing
from the analysis of the finished product. Yet its existence is no mere fig-
ment of unbridled imagination; dedicated and seasoned workers have
repeatedly asserted its reality, as when a mature V. Bertoldi titled his intro-
ductory treatise L'arte dell' etimologia. 4 Those who designate etymology as
an art seem to hint at four isolable qualities, or at any conceivable number
of their free combinations: (a) the inventive strain in the worker's mind,
which prompts him to engage in felicitous, unexpected associations of
4 This situation clashes with the Renaissance use of arte for 'practical grammar' of a
foreign tongue (P. de Alcala), including those of the New World; there arte connotes
'skill', 'training', i.e., a body of knowledge and a measure of thoroughly communicable
deftness-in contrast to the nimbus of uniqueness surrounding the preponderantly
modem concept of art. Artistry and artfulness in etymology are distinct from the artistic
pose, which may serve as an excuse for pyrotechnics or licentiousness; cf. E. Gamillscheg,
"Zur Methodik der etymologischen Forschung", Zeitschrift /iir franzosische Sprache
und Literatur L (1927), 216-298.
 ETYMOLOGY AND GENERAL LINGUISTICS 203
isolated facts or to fill successfully a gaping lacuna in the available infor-
mation; (b) a certain finesse and elasticity in bringing to bear on an
etymological problem-successively or conjointly-very disparate analyses
(phonological, grammatical, semantic), all of them within the realm of
linguistics, though at varying distances from its core; (c) the readiness, of
late increasingly rare among tone-setting linguists, to balance linguistic
evidence against an extraordinarily wide range of complementary material
-so wide as to require an encyclopedic range of curiosity and even of
expertise. This operation demands not only erudition, but also delicacy of
touch, since the number and the proportion of differently labeled exhibits
vary from case to case; (d) a flair for selecting the ideal moment to halt the
accumulation of raw data and the proliferation of digressions, also for the
calculated risk in announcing, at a chosen turn, the preliminary or final
results of the inquiry-a flair reminiscent of the talented writer's ability to
stop elaboration at the right juncture, or the experienced visual artist's
knack for refraining from, say, obtrusive overpainting. (Note that, while
there may be some affinity between musicianship and alertness to the
phonic, especially melodic, features of speech, imagery impinges on
language nowhere with nearly the same force as in those strata laid bare by
etymology and by metaphorics.)

7. Etymology and the changing climates of linguistic research. The in-

tensity with which etymology is cultivated may, then, be expected to depend
on the general climate of linguistic research in a given environment. It
seems almost platitudinous to state that a broad proclivity to historicism
(whether romantically tinged, as in the early 19th century, or adorned with
the trappings of evolutionary theory, as after Darwin) fundamentally
favors etymology, whereas a heightened concern with description, specific-
ally with distribution, at the expense of diachronic probings, tends to
retard it. The reality is far more complicate'd, since the peaks of historical
grammar and those of etymological exploration rarely coincide.
Granted the impact of Zeitgeist, one ventures to predict, for our own
time, a dual effect on etymology of that mathematical styling which has in
recent years become a hallmark of the social sciences at their most ambi-
tious: attempts to press etymological research into the mould of mathe-
matical formulae (the positive reaction to the new current) are periodically
balanced by spells of completely reckless impressionism playing havoc with
those ingredients of rationalism which have so far given etymology a
semblance of scientific respectability. For a telling instance of such statis-
tical and mathematical inroads cf. A. S. C. Ross's booklet Etymology, with
Especial Reference to English, and its rather trenchant appraisal by W. P.
204 YAKOV MALKIEL

Lehmann (Language XXXV [1950], 351-353). Some Continental journals

still abound with examples of the countertendency, the playful attitude
toward the study of lexical origins, almost on the level of genteel
entertainment.
Structural linguistics and etymology are not incompatible in their
logical foundations; but the emotional subsoils that have, at different
periods and in different places, nourished the growth of each seem indeed
to have tended to make them mutually exclusive. One thinks of several
linguists who have distinguished themselves in structural analysis (brought
to bear, e.g., on comparative Indo-European, on Slavic, on Romance, on
Dravidian, on American Indian) and who have, at the same time, refused
to turn their backs on etymology. Their data have been very neatly filtered
through stringent phonemic analysis.s But there has, to my knowledge,
never occurred a true fusion, a complete integration of the two disparate
skills, or, by way of alternative, any clear delimitation. In extreme cases
one is reminded of those by no means uncommon scientists who in their
spare time are virtuoso violinists or almost professional portrait-painters.

8. Etymological universals? Can one legitimately speak of universals in

etymology? In such a context, this term would not, of course, mean 're-
current patterns of changes in meaning', a problem which has its place in
diachronic semantics; still less 'patterns of change in form'. More defen-
sible would be the use of universal in reference to high predictability of
lexical contamination, or of deflection from regular phonetic change in
response to sound symbolism, playfulness, "expressivism", and the like-
designations which directly affect the very kernel of etymological research;
still, one might be in doubt whether such matters do not come more appro-
priately under the heading "linguistic change". An irreducibly etymological
universal, on the other hand, would be a recurrent degree of genetic
transparency.
Thus, in numerous languages the-parallel or discrepant-words for
s Some notable examples of continued etymological curiosity on the part of scholars
committed to structuralism: R. Jakobson's substantial review here, in several instal-
ments, of M. Vasmer's Russisches Etymologisches Worterbuch and his earlier lexical
studies buttressing his dating of the Igor' Song; M. B. Emeneau 's Dravidian Etymological
Dictionary (1961, in cooperation with T. Burrow), as a sequel to his descriptive research
in Kota and Kolami; M. R. Haas's articles on the Proto-Gulf and Proto-Hokan-
Coahuiltecan words for 'water', International Journal of American Linguistics XVIJ
(1951), 71-79 (cf. the comment ibid. XXIII, 7) and University of California Publications
in Linguistics X (1954), 57-62; W. Bright's recent Indianistic study in Californian animals
of acculturation (University of California Publications in Linguistics IV:4), cf. Romance
Philology XIV (19~1), 360-361; and the astonishing diapason of E. Benveniste's
inquiries, reminiscent of Sapir's.
 ETYMOLOGY AND GENERAL LINGUISTICS 205
'boy' and 'girl' show not only an unusual rate of attrition (incidentally, for
slightly different reasons), but also an alarming degree of etymological
intricacy: Engl. boy, Fr. gar(:on, Sp. muchacho, OSp. mancebo, Ptg. mo(:o
and menino, to adduce a few examples out of hundreds, 6 have all long
baffled seasoned etymologists and involve, at least in some recalcitrant de-
tails, problems still unsolved. Another scaling of this kind may be carried
out with the names of domestic animals: in the Romance domain those of
the females ('nanny goat', 'cow', 'mare', etc.) display an incomparably
higher rate of preservation and concurrent etymological translucency than
those of the corresponding males (both reproductive and castrated) and of
the young. 7 In either case the socio-cultural matrix is to a higher degree
responsible-for this special lack of transparency than are purely linguistic
conditions. Any parallel appeal to interjections may be summarily dis-
missed with the remark that these take us to the very periphery of grammar
and lexicon. A distinctly more promising avenue of approach is Meillet's
observation-based on inspection of paleo-IE material-to the effect that
adjectives offer a far greater proportion of etymological complications
than, say, equally abstract verbs. The Romance languages confirm this
suspicion, and if it receives substantial support from other sources, we may
come close to identifying, at least, one tendential etymological universal.

9. Typology of etymological studies. A not unwelcome touch of strin-

gency can be added to etymological research through use of the typological
approach. Thus, by resorting to a technique of analysis known as "typo-
logy of the genre" one can resolve the overwhelming majority of etynio-
logical studies, kaleidoscopic as they appear at first glance, into certain

6 Aside from A. Sperber's pilot study (1911), the key monograph, especially for
Gallo-Romimce and Italian, remains I. Pauli, 'Enfant', 'garc:on', 'fille' dans les langues
romanes (Lund, 1919). Though essentially a "travail de patience", it provoked weighty
reviews, identifiable through Hall's bibliography, by experts (A. Wallenskold and 0. J.
Tallgren [-Tuulio], E. Tappolet, L. Spitzer, A. Castro, G. Rohlfs, W. von Wartburg,
L. Jordan); cf. Rohlfs, AR VIII (1924), 161-166. On boy see E. J. Dobson, Medium
Aevum IX (1940), 121-154. There has recently been no uncertainty about the Frankish
provenience of garc:on, but authorities disagree as to the specific base. OSp. Ptg. moc:o
bas been another apple of discord; note that J. Corominas, Diccionario critico etimo-
IOgico de Ia lengua castel/ana III [1956], 463b-465b, seriously questions the widely
accepted etymon musteu 'musty, fresh' and toys with reverting to Schuchardt, Baist, and
Garcia de Diego's minority view. Possibly the latest statement on the suffix of much-,
orig. moch-acho is found in Language XXXV (1959), 215-224, esp. fn. 75. Sp. nino and
Ptg. menino may be congeners, but what of me-? It. ragazzo need not be onomatopreic,
but certainly resists analysis.
7 One thinks of Sp. chivo 'kid', garanon 'stud jackass' (Amer. 'stallion'), jato 'calf';
IL becco, and the like.
206 YAKOV MALKIEL

more or less fixed categories. One such experiment (International Journal of

American Linguistics XXIII [1957], 1-17) was conducted as follows: The
multitude of possible fruitful approaches were broken down according to
three major criteria: (a) the varying scope of the inquiries; (b) the kind and
amount of material adduced as evidence; (c) the inherent degree of sim-
plicity of each problem at issue (cf., in the present article, the statement on
transparency in the preceding section). The order of these three criteria may
be freely inverted, and the possibility of introducing others of equal rank
remains open. Application of the first criterion allows one to set off:
(a) the entry in an etymological dictionary (of variable size, specificness,
styling, etc.); (b) the short etymological note (a bit of independent glean-
ing, a shred of supplementary information, a trial balloon), (c) an item
culled from a historical grammar's mosaic of etymological equations;
(d) the by-product of some such philological venture as the edition of a text
studded with lexical difficulties; (e) a middle-sized self-contained etymo-
logical study; (f) a major article or full-blown monograph, which in turn
may alternatively stress (a) the origin and early development of a single
word, with emphasis on the fluctuating semantic ambit; (fJ) the specific
evolutionary anomaly identified as the crux of a unique difficulty; (y) the
clustering of several etymological problems, typically around a note-
worthy historical situation; (S) the dyadic or triadic arrangement of ety-
mological problems as a result of such purely linguistic situations as lexical
polarization and serialization (for additional data see Archivum Linguisti-
cum IX [1957], 79-113, esp. 103 ff., and X [1958], 1-36); (e) th~ impact of
homonymy; m the array of (near-)synonyms, especially that kind of cross-
dialectal synonymies which reached its crest in the inter bella variety of
"onomasiology" (cf. B. Quadri's survey of 1952), entirely unrelated to
onomastics as currently understood.
One can establish, in comparable detail, some kind of scale or gradient
of etymological researches on the basis of the two remaining criteria. The
conceivable crowning achievement of this approach-unfortunately, omit-
ted from the tentative survey five years ago-might be a complete integra-
tion of the three analyses: Does there (or, at least, should there ideally)
exist a close correlation, describable in specific terms, between scope,
material, and degree of intricacy? Such a final balance-sheet would not only
cut a path through the maze of the etymological output, but serve as a
frame of reference (by way of encouragement and of deterrent) for future
planning.

10. Typology of problems and of solutions. Distinctly richer in prospects

is a typology of etymological problems and of their successive analyses;
 ETYMOLOGY AND GENERAL LINGUISTICS 207
the sketchy presentation in Lingua V: 3 (1956), 224-252, with its emphasis
on a single intrinsic discrepancy ("uniqueness vs. complexity"), provides
little more than a foretaste of this classification. Any protracted etymo-
logical debate can be epitomized in, at least, two fashions:
(a) In straight annalistic manner, as a zigzagging narrative, with full
attention to each new conjecture and to each fresh facet of an old con-
jecture newly championed as well as to the miscellaneous reactions such
proposals elicit, until the problem has either been solved through a con-
sensus, or, if the discussion grows sterile, been shelved pending the dis-
covery of some new decisive piece of evidence. This approach ordinarily
takes notice of the "human element", the drama pervading the debate, and
is thus at the farthest conceivable remove from austere scientific styling;s
(b) With the material and the matching ideas grouped in more abstract,
analytical fashion around the major solutions-provided one rigorously
subordinates all interludes. The bases advocated may, but need not, follow
a strictly chronological line; if they do, that line is either a string of the
exact dates of formal scholarly identifications, or a sequence of the ap-
proximate dates when assumed bases emerged in actual speech. As an
alternative to historicism in either garb, the hypotheses could be arrayed
on the basis of source language, word family, derivational structure, or
semantic background. 9
From this less narrative, more interpretive treatment one arrives (as the
1956 article failed to make sufficiently clear), by deliberate trimming-
omission of such conjectures as are of merely antiquarian interest or mark
a step backward-, at the actual "nucleus of the problem", frequently a
restatement of some irreducible dilemma.1° Essentially, this sifting of
opinions and narrowing-down of choices represents a preliminary analysis,
through injection of value judgments. Further gradual schematization
allows the analyst so to simplify a highly complex state of affairs as to have
on hand a mere residual formula symbolizing the farthest advance short of
the so far unachievable break-through. He can then proceed to categorize
sa. my Luso-Hispanic studies of albricias '(reward for) good news', Studies in
Philology XLIII (1946), 498-521, esp. 499-504; asperiega, Carol. sperauca 'species of
apple', Philological Quarterly XXVIII (1949), 294-311, esp. 295-298, 308-310; despedir
'to dismiss', refl. 'ask leave', University of California Publications in Linguistics XI (1954),
40-42, 155-157; lozano 'proud, exuberant, verdant' and Ptg. trigar-se 'to hasten', ibid.,
1:7, 244-267, 283-288.
9 Cf. my studies of Sp. lerdo 'dull, slow', Philological Quarterly XXV (1946),
289-302, esp. 289-292, and OSp. maznar 'to knead', Modern Language Review XLIX
(1954), 322-330, esp. 323-325.
1o Cf. my study of OSp. cuer ~ coraron 'heart' (the latter orig. *'heartburn') in Bulletin
hispanique LX (1958), 180-207, 327-363, esp. 195-197.
208 YAKOV MALKIEL

the various types of impasses: genuine alternatives, specious alternatives,

instances of admittedly complete ignorance, examples of a single pre-
ference still insufficiently substantiated and of hesitation between three or
more partially acceptable solutions.
It seems equally rewarding to build up a separate typology of ultimately
successful solutions, distinguishing between (a) clarifications achieved
through outside help (e.g., through discovery of unsuspected raw facts),
(b) satisfactory compromises between earlier not quite acceptable pro-
posals, and (c) fruitful mergers of two different techniques of analysis, each
inadequate if used in isolation.

11. The graded approach. Haphazard as may be the sequence of the

actual flashes of etymological thought, one need not report them in such
capricious succession; nor is it customary to proceed quite so impatiently.
However, the currently favored alternative to haphazardness, namely
assembling etymologies in alphabetical order, has equally limited useful-
ness: Though it facilitates casual consultation, it fails to highlight the actual
drama of etymological probings and, in particular, to identify the fluctuat-
ing front-line. For all its unquestioned practicality as a reference tool, the
etymological dictionary, viewed as a scholarly enterprise, reminds one of
the phonological section of an old-style historical grammar, in which the
individual sound correspondences were established tidily enough to enable
the philological reader to classify cogently any manuscript at issue along
the two axes of time and space, but which consistently neglected the inter-
action of the shifts expertly identified. One way of presenting etymologies
in a significant sequence likely to stimulate further research is to arrange
them roughly in a line of increasing complexity-on the basis, as it were, of
the presumable number of unknowns; an appended alphabetic index may
then provide the necessary references. Such an experimental array, sug-
gested in Word VI (1950), 42-69, was not meant to be binding on anyone,
not even on its proponent, so that its acrimonious rejection by the ranking
advocate of free-wheeling practice (L. Spitzer in Romanische Forschungen
LXXII [1950], 227-234) was quite unjustified. In reality, the opposite pole
to the policy of gradualism is represented by a deliberate concentration on
the most rebellious, hence titillating problems, as when J. Hubschmid tends
to by-pass the Latin layer of Romance in favor of the vastly less accessible
substrata! bases; for criticism, see my review of two monographs of his in
Language, Vol. XXXVIII.

12. Etymology and diachronic structure. The cornerstone of the entire

edifice of etymological theory is the paradox that this area of knowledge,
 ETYMOLOGY AND GENERAL LINGUISTICS 209

though an inalienable part of lexicology, owes its prime distinction to its

conspicuously close ties with diachronic grammar. While in the descriptive
perspective structure and lexicon, despite certain points of contact and
even an occasional overlap, lend themselves to neat separation, the lan-
guage historian in his daily work finds them inextricably interwoven. This
intimate relation is not the least striking peculiarity of all manner of glotto-
dynamics, as against all, or nearly all, static approaches to language.
Numerous specific problems of theory and practice flow from this general
situation.

12.1. Estrangement between etymology and grammar. No historical

grammar can boast any measure of usefulness unless its smallest consti-
tuent elements-the etymological equations-have been painstakingly
refurbished. In the classic age of historical linguistics this truism posed no
serious problem, since the same experts worked-often almost simul-
taneously-on grammatical and on lexical projects; witness Grimm,
Meillet, Ernout; or, for that matter, Diez and Meyer-Li.ibke. In recent
decades, however, the liaison has slackened, with the result that the gram-
marian may deem himself exempt from the obligation of keeping abreast of
etymological advances. Among the preponderantly severe reviews of M.
Regula's ill-fated historical grammar of French (1955-1956) K. Baldinger's
was even more enlightening than those emanating from structuralistic
quarters, disclosing as it did that the book's author, not only conservative
but, for good measure, slipshod, was sorely vulnerable on his own ground
(see Zeitschrift fiir Phonetik, XI [1958], 282-288; cf. Romance Philology
XIV [1960-1961], 361-362).
The harmful gap between etymology and grammar began to widen
thirty to forty years ago through a fatal coincidence: While leading lexico-
logists apparently felt that old-style grammar, especially if applied to the
more familiar languages, had nearly exhAusted its possibilities (which was
almost true), that there was little point in further experimenting with
grammatical material except possibly in derivation and syntax (which was
patently untrue), and that new factual (i.e., in this context, etymological)
insights alone could lead to eventual progress (at best, a one-sided decision),
the die-hard grammarians, attacking preferably languages hitherto un-
surveyed, were quick to establish the reverse scale of values and priorities.

12.2. The varying impact of etymology on branches of grammar. The dia-

chronic interpenetration of etymology and grammar is not evenly distri-
buted over the entire expanse of the latter. It reaches its maximum intensity
in certain divisions of phonology and in that broad province of morphology
210 YAKOV MALKIEL

which has been variously labeled as derivation or word-formation, embrac-

ing both affixation and composition. It is less apparent throughout in-
flection-though it clearly matters to the student of Spanish conjugation
whether OSp. trofir 'to pass' reflects triiducere (Nueva revista de .filologfa
hispanica X [1956], 385-393), how reliable is the connection with OSp.
defir, Ptg. de(s)cer 'to descend'. claimed variously for desidere, decidere,
deicere, discidere, and discedere (cf. Lingua V, 229-230), or just what is the
protype of asir 'to grasp' and (de)rretir (Ptg. -er) 'to thaw, melt'-given the
severely limited vitality of the local -er and -ir classes. Syntax and etymo-
logy share fewer interests, though one can point out an occasional en-
croachment. Thus, J. Jud's brilliant demonstration, with the aid of a
Raeto-Romance congener, that OFr. estovoir 'to be necessary' derives, in
the last analysis, from est opus (Vox Romanica IX [1946-47], 29-56), the
cleavage of Lat. fallere 'to deceive' into Fr. falloir 'to be necessary' and
faillir 'to err, be on the point of' (beside Sp. fall-ecer, orig. -ir 'to run out,
decease') and the transformation of capere 'to seize, catch' into Sp. Ptg.
caber 'to be contained in', cf. laceriire 'to tear'> OSp. laz(d)rar 'to suffer
hardships', condire 'to season'> Sp. cundir 'to spread, swell, multiply'
(intr.), all five bear on such fundamentally syntactic problems as personal
vs. impersonal or transitive vs. intransitive construction in provincial Latin
and Romance.
The link with historical semantics is so self-explanatory as to require no
documentation-a decision possibly the more welcome as the field rules for
the formulation of semantic change are in process of reorganization; cf. in
particular E. Benveniste, Word X (1954), 251-264. This link must be
scrupulously distinguished from etymology's miscellaneous intrusions
upon the territory of synchronic semantics-via either folk etymology or,
less familiar, the artistic reinforcement of fading etymological connections
(S. Ullmann, Precis de semantique franfaise [Berne, 1952], pp. 115-120).
An additional line of inquiry, dear to E. Lerch and programmatically
stated by G. Gougenheim, would be the systematic search for erroneous
learned adjudication in matters etymological, u a search presupposing an
antiquarian curiosity akin to that which presides over mosaic reconstruc-
tions of etymological debates (see Section 10, above). At the intersection
of the semanticist's and the antiquarian's lines of thought one might place
the procedure of a poet so learned as to have tendentially included, in his
quest for elegant ambivalence, the favored key-words' etymological mean-
ing as one of their many splendors, but neither so well-informed as to have
selected a truly up-to-date guide to word origins, nor technically expert
enough to have made forceful decisions of his own in the face of erudite
u "La fausse etymologie savante", Romance Philology I (1947-48), 277-286.
 ETYMOLOGY AND GENERAL LINGUISTICS 211
controversy. Paul Valery is rumored to have been one such poet, ever in the
throes of etymological anguish;12 if this report is correct, his hidden
etymologizing would be on a par with Haubert's Carthaginian archeology,
Rim baud's evocation of tropical South America, and Tolstoy's logistics.

12.3. Need for continued momentum of etymological research. Those

linguists who view etymology with a jaundiced eye sometimes question its
usefulness beyond the collection of a small sampling of equations needed
to establish the all-important phonetic correspondences. This indifference
may be countered with two arguments. First, the edifice of phonology can
be improved both through refinement of method (the currently fashionable
approach) and through continued expansion of factual knowledge (here
etymology acts as a powerful cutting wedge which the trained linguist
deftly drives into layers of unknown material). Second and more important,
even the staunchest "regularists" admit that an original phonetic corres-
pondence-inferrable, e.g., from the total sound pattern-may have been
locally overlaid on a considerable scale, sometimes to the point of shrink-
age.I3 If such an abnormal situation-by no means infrequent-has crys-
tallized, then the last haul of painstakingly established etymologies may
constitute, not some unexciting aftermath merely confirming or further
specifying earlier findings, but the first direct clue to an earlier radically
deviant stage so far adumbrated only through circumstantial evidence.

12.4. Overlap between grammar and etymology. Between the domains of

historical grammar and of etymology there exists a sizable, possibly in-
eradicable overlap. Decisions as to whether a set of problems merits dis-
cussion under the one or the other head affect the economy of research and
thus become an essential part of long-range planning (cf. Romance
Philology VIII [1954-1955], 187-208). There are two reasons for this
tangle. '
On the one hand, to the extent that interpretive historical grammar
moves away from mere cataloguing and that numerous minute facts refuse
to fall into the vitally important broad patterns, the residue created by this
hierarchization tends to form a kind of no man's land. Countless phono-
logical details thus dislodged from their former niches in footnotes and
12 He was ill-advised enough, I understand, to select Cledat as his etymological
mentor. Rilke, in his correspondence, admitted consulting Grimm's DWfortranslation
purposes, but his lexicon (unlike George's and particularly Borchardt's) remained con-
sistently free from deliberate archaisms of form, let alone allusions to obsolete meanings.
ll One relentless champion, almost twenty years, has been R. A. Hall, Jr.; see, in
particular, Italica XXIII (1946), 31-32 (fn. 5) and Romance Philology XV:3 (1961-62).
212 Y AKOV MALKIEL

small-print sections (as still predominant in Grandgent's From Latin to

Italian [1927]) are of potentially greatest interest to the etymologist. Thus,
the progressive author of a historical Romance grammar, to avoid miring
in irrelevant details, may fully list the sharply silhouetted development of
stressed Latin vowels-few and each copiously represented-but merely
exemplify prevalent trends where the extraordinary variety of secondary
consonant clusters is involved. True, of those newly learned by speakers
some were neatly predictable from the total pattern (e.g., -xtr- [str] as in
OSp. yxtre 'I shall go out', from exir, cf. -mbr-, -ndr-, etc.). But other ex-
amples of secondary contiguity (say, OSp. -zd-, -zl-, -zr-, -zt-) display both
an unusual configuration and an extremely low incidence; speakers,
apparently reluctant to acquire a neuro-muscular skill of such scant use-
fulness, resorted to an astounding fange of devices to lighten their burden, 14
with the result that regularity of sound change reached its nadir.
On the other hand, the etymologist again and again encounters minor
but obnoxious stumbling blocks-e.g., puzzling developments of sounds or
affixes-for which he cannot, with a clear conscience, refer his reader to
any standard historical grammar. What is he to do? He may limit himself
to hinting briefly at a similar difficulty identified in another word-a light-
ness of touch which leaves the problem slightly widened, but still unsolved
(this practice seriously vitiates Spitzer's conjectures); or he may inflate his
dictionary, if such is the form of his project, with long-winded phono-
logical discussions under the one favored entry and later provide a few
cross-references, a solution adopted by Corominas (1954-1957) and
amounting to a scattered grammar concealed within a dictionary.
Gillieron's idea of supplying a string of supplements or excursuses, as he
did in his "genealogical" monograph on the Gallo-Romance names of the
bee (1918), may, on balance, be more felicitous, except that such discus-
sions need not be relegated to the very end: all that matters is to set them
off with maximum tidiness, to ease the reader's strain and to catch the eye
of some future grammarian. 15 This preference implies that the proper

14 For a detailed discussion of Lat. -cer- > OSp. -zr-, apropos of lacerare 'to tear to
pieces' and mticerare 'to soak', see Nueva revista de filologia hispdnica VI (1952), 209-
276, and Modern Language Review XLIX (1954), 322-330. On the importance of relative
yield I find myself in agreement with Martinet. There exists, of course, no direct ratio
between yield and regularity, since the total sound pattern exercises its share of control-
ling influence, occasionally making an infrequent sound shift astonishingly firm. But
given equality of all conditions (including qualitative suitability), a sound shift common
in terms of lexical representation and of incidence stands a good chance of scoring a
high degree of regularity.
15 In retrospect I find my own earlier performance heavy-handed, cf. the long para.
graphs (a) on patterns of vowel dissimilation in the study of OSp. re-, sa-codir (Hispanic
 ETYMOLOGY AND GENERAL LINGUISTICS 213

place for microscopic etymological inspection should be the article or the

monograph rather than the dictionary, which ought to be returned to its
proper size and role of a mere inventory of succinctly classified and docu-
mented forms, and an accurate guide to pertinent discussions.

12.5. 'Weak' phonetic change, sporadic sound shift, lexical contamination.

The relation between etymology and phonology may be expressed in terms
far more intimate than those suggested by tactical or strategic considera-
tions. There is a strong possibility that the very frequency or degree of
etymological opaqueness may to some extent depend on internal phono-
logical conditions. Suppose some sound disappears consistently in the
west. survives almost invariably in the east, and exhibits an erratic record
of persistence in the center of a given territory, as is, e.g., true -except for
one detail-of Latin primary intervocalic -d- in the Iberian peninsula (short
of the Catalan zone; see Language XXXVI [1960], 284-290). Presumably
old dialect mixture (due to successive political reapportionments of the
area?) accounts for the instability in the Center. Be that as it may, pre-
dictability is so low in the transitional territory as to entitle us, irrespective
of the historical explanations advanced, to posit a 'weak' phonetic law.
One may now plausibly make this surmise: (a) Such a state of low pre-
dictability of regular phonetic change, aside from actually entailing specific
vacillation (cruo,..,crudo, nfo,..,nido, etc.), creates, in regard to this feature,
a general climate of insecurity which (b) invites an extra-heavy proportion
of lexical blends (associative interference), and (c) concurrently stimulates,
outside the pale of regular phonetic change, those more or less latent sound
shifts which have long been known as "sporadic" or "spontaneous" (haplo-
logy, metathesis, dissimilation other than in contact, etc.).16 Clearly, each
of these three discrete forces viewed in isolation would of itself tend to
obscure a word's provenience. If a statistical demonstration can some day

Review XIV [1946], 133-135) and (b) on the alternation of OSp. [dz]-[z] apropos of
cosecha (Language XXIII [1947], 389-398, esp. 393-397). For later examples of chapters
or sections so clearly marked off as to make them easily transferrable to grammatical
terrain, see the discussion of s(s)-~ (210-222) in the paper on OSp. assechar 'to stalk'
(Hispanic Review XVII [1949], 183-232) and comparably placed statements on the
wavering between per-, por-, pro-, and pre- (Romance Philology III [1949], 27-72, esp.
61-67) and on the sources of the cluster -ld- (Est. Menendez Pidal, I [1950], 91-124,
esp. 102-121).
Ib For a more fully developed preliminary statement see my forthcoming contribu-
tions to (a) the A. W. de Groot Testimonial ("Weak Phonetic Change, Spontaneous
Sound Shift, Lexical Contamination"), and (b) the Melanges M. Bataillon, a Franco-
Brazilian venture ("Etimologia y cambio fonetico debil: Ia trayectoria iberorromanica
de medicus, medicina, mediciimen").
214 YAKOV MALKIEL

actually gauge the mutual attraction and consequent reinforcement be-

tween the three, the integration of etymology with the core disciplines of
historical grammar will have taken a major step forward.
What increases the plausibility of such interaction is the fact that it provides the
missing link which at once straightens out numerous, at first glance disconnected,
difficulties. One is predisposed to link Sp. peldaflo 'step (of a staircase)' to pecili11eu
'pertaining to the foot', but the epenthesis of -1-, viewed out of context, is baffling
(this circumstance makes the analysis in Archivio Glottologico Italiano XXXVI [1951 ],
49-74 seem incomplete, despite the profusion of details). The link between OSp.
vedegambre 'poison' and mediciimine appears at first glance self-evident, but why has
•medegambre left no easily detectable traces, while among the syncopated variants
megambre visibly flanked vegambre 'id.' (free alternation of initiallabials)? The closer
one inspects Sp. calavera 'skull', fig. 'madcap', the more one is perplexed as to which
of its two obvious ingredients, cadiiuer 'carcass' or caluiiria 'scalp', constitutes the
irreducible base, and which merely the contaminator. A dozen or more such con"er·
gent remifications of problems arouse one's suspicion that the weak growth of the -d·
was a prime factor in activating the two other categories of shifts.

12.6. Non-cultural concomitants to learned transmission. Whether or not

the application of this principle may lead to sensational discoveries, its
chief merit so far lies in strengthening our grasp of fundamentals. It takes
only a modest training to discern the 'learned' character of Sp. dulce 'sweet'
as against Ptg. doce, It. dolce, Fr. doux, etc., all of them outgrowths of Lat.
dulce; but a cogent account of this regional departure requires sophistica-
tion. To be sure, dulcis, -e, with its rich spectrum of connotations, occupied
a place of honor in Church Latin, but why should that common matrix of
Western medieval culture have at this point unequally influenced Romance
vernaculars so closely akin? The discovery of inconspicuous Old Spanish
by-forms (doz, duz, dulz, etc.) and the realization that the development of
Latin syllable-final 1 precisely in Proto-Spanish was erratically tripartite
(loss,..., shift to semivowel,..., preservation), thus entailing a very weak
phonetic change, jointly alert us to the possibility of a speech community's
escape from protracted wavering through adoption of a fixed prestige
form. This therapeutic explanation echoes Gillieron's earlier contention
that French occasionally appealed to Latinisms to solve homonymic con-
flicts.

12.7. Etymology and derivation. There exists a comparably intricate set

of relationships between etymology and derivation. An unusual affix,
especially if known to have long been unproductive, may be the safest clue
to an etymological crux, cf. Sp. pendencia 'quarrel' <poenit-entia '[loud]
repentance' (Romanic Review XXXV [1944], 307-323), OSp. cans-a(n)cio,
Ptg. cans-oro 'fatigue' <quass-titi6 'break-down' ( x campstire 'to dodge')
 ETYMOLOGY AND GENERAL LINGUISTICS 215

and OSp. posfacar 'to mock', OGal. pos-faz 'mockery' <postfaciem (ridere)
'(to laugh) behind one's face' (Romance Philology III [1949-1950], 27-72).
Conversely, a paucity of tell-tale derivatives isolates the word etymologic-
ally and may critically retard its classification; in Indo-European this
liability affects most primary conjunctions and prepositions and many
pronouns (cf. Word X [1954], 265-274),17
To serve as a guidepost, the affix need not be exceptional per se; what
matters most is the aberrancy of its relation to the root morpheme. Thus,
the -a of Sp. burr-a 'she-ass', fig. 'drudge', perr-a 'bitch' neither poses a
genetic problem nor sheds any light on the disputed ancestry of burr-o,
perr-o; but mentir-a 'lie', involving the same desinence uniquely joined to
an infinitive, displays a potentially revealing pattern (Romance Philology
VI [1952-1953], 121-172; for additional examples, see Word X, 269 ff.).

12.8. Summits of grammatical and of etymological research. Since

progressively less gross phonetic correspondences are established by in-
creasingly microscopic inspection of "residues", it is natural that the most
refractory items, which invite the closest etymological examination, should
mainly claim the attention of that generation of scholars which follows
upon the creators of monumental grammatical syntheses. It is equally un-
derstandable that such a generation, not necessarily of epigones, should
tend to develop radically different study habits and gamuts of taste. In
Romance scholarship, the leading dialect geographers, through a note-
worthy twist of events half a century ago, assumed the collateral responsi-
bility for the etymological aftermath. This activity, of which they
brilliantly acquitted themselves, sensitized them to the multiple cross-
connections between (to quote Bally) language and life, but, psychologic-
ally, blunted their curiosity about structure-since their task was to react
against extremes of neogrammatical schematization and isolationism-and,
as a result, thwarted any possibility of their fruitful cooperation with the
new schools of structuralists.
17 The coexistence of significant variant forms ordinarily favors the etymologist, while
their lack impedes him (cf. Sp. Ptg. tamar 'to take'<?; Fr. tirer, It. tirare 'to draw',
Sp. Ptg. tirar 'to throw, fling, draw'< ?), though there is no dearth of richly diversified
lexemes awaiting etymological clarification (e.g., OSp. combr-, combl-uera, Ptg. com-
borra, etc. 'concubine'). Can one draw chronological conclusions from the protracted
coexistence of variants neither regionally nor semantically diversified? Some cautious
inferences as to fairly recent coinage or, at least, reshaping are perhaps admissible,
especially if one part of a given paradigm clashes with the remainder, relatively free from
such oscillations; cf. such ill-delimited syncopated futures as OSp. porre, porne,
pon(d) e, beside root morphemes of other tenses of poner leveled to the point of mono-
tony. It behooves, then, the student of inflection rather than the etymologist to conduct
the initial experiments along this not unpromising line.
216 YAKOV MALKIEL

13. Logical implications of etymology. Despite one's misgivings about the

approach to word origins via mathematics, one discerns some undeniable
contacts between logic and etymology. Here are a few random illustrations.

13.1. Ambiguities hidden in uniquely correct equations. Not only is an

etymological solution unique whatever its complexity, but it may be de-
finitively correct despite our temporary or permanent inability to give un-
equivocal account of one or more evolutionary stage(s). Thus, the genetic
link between, on the one hand, pectoriile 'pertaining to the chest' and, on
the other, archaic Sp. peitral> Old, Class., dial. petral> Mod. pretal
'(horse's) breastband' is unassailably certain-the eventual metathesis
being due to the influence of descendants of the premere family (BICC
IX [1953-1955], 1-135)-thougk it remains arguable whether the latest
Latin and the earliest recorded Central Hispanic form can be most effec-
tively bridged by *peitoral (as was actually used in Portuguese) or by
*peitrale, or by both, according to the area at issue. If the last-mentioned
possibility holds water, then a single word would have split into two which
ultimately coalesced.

13.2. Diverse categories of hypothetical bases and intermediate forms. In

operating with hypothetical bases, scholars have been traditionally satisfied
with the rough distinction between recorded and unrecorded forms. Under
a single undifferentiated symbol, the asterisk, we tend to subsume (a) items
of whose existence we are so firmly convinced as to deem their absence from
the record purely accidental (e.g., Lat. *credentia 'belief',judgingfrom Ptg.
crens;a, Sp. creencia, Fr. croyance, Rum. credinfii, etc. and against the
background of fervent faith characteristic of early Christianity; starred
intermediate forms are very frequent in reconstruction) and (b) items whose
existence we expressly deny. The latter comprise, inter alia, (a) ideal out-
comes of phonological drift-results which some such interference as
devious channeling (e.g., learned transmission), morphological leveling, in-
ternal lexical association, or diffusion prevented from materializing (niiru
'daughter-in-law'> It. nuora, not *nuro); and ([3) deliberately misleading
forms used as a pedagogical foil against which the true development stands
out the more sharply (manu> Fr. main, not *men). A logically preciser
notation may distinguish between these two, or even three, possibilities
(cf. Word VI, 49-57).18

18 E.g., one asterisk placed at different levels (a subscript star was widely favored in
the mid-19th century), or varying constellations of asterisks (*, **, etc.). While de-
scriptivists have gone overboard in their enthusiasm for newly devised signs and unusual
fonts, language historians have displayed undue restraint. True, large capitals arc used
 ETYMOLOGY AND GENERAL LINGUISTICS 217
13.3. New auxiliary constructs: generalizers and standardizers. To relieve
the pressure on the asterisk, and to establish a rapport with the symbolic
logicians' uninterpreted forms,t9 one may introduce other constructs.
J. Jud, a superb practitioner but, unfortunately, no theorist, launched half-
bracketed forms to mark, if one may revert to his own and his peers'
terminology, lexical types. Upon closer inspection such forms tum out to
function (a) as generalizers, bracketing in non-committal fashion several
closely related forms; and (b) as standardizers, replacing one highly special
form, not immediately transparent to the outsider, by a variant more
readily assimilable, though, strictly, non-existent.
Generalization may be an advantage in dealing (a) with cruxes, as when
one subsumes under common denominators like ra[(/)arel, rtirarel,
rtomarel, rtropare1 scattered dialect forms without committing oneself as
to specific source language, exact primary meaning, any minor detail of
form (rtomiire1 or ru:imiire1?), or combinations of these three features.2o It
is further justified as (ft) a makeshift designation of a lexical nucleus still
unanalyzed but provisionally classed as unified, if one leaves open the
possibility of a homonymic tangle (Sp. rcach-1: Corominas, Revista de
filogfa hispanica VI [1944], 33-34; rpech-1: Language XXVIII [1952], 299-
338; cf. also Hispanic Review XXI [1953], 20-36, 120-134). Operating
with such uninterpreted bases, the analyst voluntarily suspends judgment
on points he deems inessential in this particular context, and thus,
besides saving time and effort, avoids the risk of bogging down; he also
contracts a debt, because after isolating and quickly by-passing an area
of indeterminacy he places himself under the obligation of later redeeming
the mortgage.
As a standardizer, the half-bracketed form provides the ideal counter-
part, in the standard language, of obscure dialect forms, even if the word
to set off epigraphic material, small capitals may signal Latin (and other ancestral) bases
of Romance words, roman boldface marks transliteration, as in Mozarabic. Why not
use italic boldface for lexical items culled from glosses, and italic small capitals for
numismatic data? We should further restrict the signs > < to strictly phonetic change,
introduce-++- for derivation, also devise a set of differently shaped arrows for diffusion,
and select symbols less crude and more nuanced than + or x for associative interference.
19 See I. M. Copi, "Artificial Languages", in P. Henle, ed., Language, Thought, and
Culture (Ann Arbor, 1958), pp. 96--120, esp. 102-103; cf. International Journal of
American Linguistics XXV (1959), 131.
20 This use simplifies operations with regional lexical types found in areas of highly
diversified dialect speech, such as Franco-Proven~ and Raeto-Romance. In all likeli-
hood the divergent minor features mark a recent overgrowth of narrowly local innova-
tions, through which the reconstructionist must blaze a trail. To the historian, then, the
half-bracketed form symbolizes a provincial prototype; but synchronically it lends
service as some kind of master-key.
218 YAKOV MALKIEL

at issue happens to have become extinct in the standard or to have at

all times been alien to it. Thus, though devidoir ( < OFr. des-vuid-oir,
-eour) 'skein-winding reel' survives only on the patois level, we may use a
legitimate shortcut by contending that at PP. 459 and 478 of Gillieron's
atlas map LdevidoirJ and Ltravoui/J overlap. 21 If a tool known in French
as tourn-ette corresponds to Bearn. Ltourn-etJ and, in Aveyron, Lozere,
Haute-Loire, Ardeche, to LtourJ, the use of "idealized" rather than actual
forms involves an abstraction helping writer and reader to jot down just
the one distinctive feature of immediate interest in, say, a suffix study and
to disregard all irrelevancies.
14. Sound symbolism, onomatopeia, expressivism. These three key terms
presumably mark that aspect of etymological research which is fraught
with the greatest number of intrinsic difficulties (sometimes called 'in-
tangibles'), increased by a heavy accumulation of haziness on the part of
generations of analysts. Stringent scholars, thoroughly aware of these
forces (Sapir hinted at playfulness in language [1921], Bloomfield's manual
dissects symbolic words: §14.9), have tended to shirk, sidetrack, or post-
pone their discussion, while imaginative etymologists, from Schuchardt on,
may conversely have sinned on the side of overindulgence. This entire range
of problems clamors for cool-headed reexamination, with sharp distinction
between absolute and relative sound symbolism (the former based on
physiological realities, hence cross-linguistic, as in Jespersen's famous
treatment of [i] [1922; cf. Nos. 312 and 316 of his Bibliography]; the latter
flexibly geared to specific sound systems), also between primary and
secondary expressivism, i.e., a recognizable design which a word may
acquire at different stages of its growth.22 If impressionism and circular
reasoning are to be avoided, "expressivism" had best be independently
attacked from two angles: (a) the psycholinguistic analysis-possibly with
laboratory apparatus-of such sensations as provoke measurably strong
reactions (gaudy colors, shrill, loud, or intermittent sounds, circular or
zigzagging lines, exceedingly thin or thick bodies), in conjunction with
their lexical correlates, such as those surveyed in Jaberg's masterly study
of the Portuguese designations of the 'swing' (Revista portuguesa de
fi/o/ogia I [1947], 1--44); (b) a purely linguistic analysis of varying partial
deviations from completely non-committal forms {the Geneva School's
21 For details see C. H. Livingston's well-known monograph (1957) and its discussion
in Romance Philology XII (1958-59), 262-282. (It is tempting to go beyond Jud and use
the complementary symbols L J for this standardizing function.)
22 a. the contrasting semantic ambits of Sp. duendo and Ptg. dondo, both regularly
developed from Sp. domitu 'tamed' (Homen. A. M. Huntington [Wellesley, 1952],
pp. 361-392).
 ETYMOLOGY AND GENERAL LINGUISTICS 219

"arbitraire du signe") via suggestive vocalic gamuts, reduplication of a

word's core segments, and the like-provided such geometric configura-
tion fulfills no grammatical requirement (cf. Revue de linguistique romane
XXIV [1960], 201-253, esp. 221-233). Psycholinguistics may some day also
buttress the unduly discredited domains of onomatopreia (as in evocations
of animal cries) and interjection.

15. Concluding remarks. Aside from its temporarily forfeited controlling

position within linguistics, etymology, deftly handled, is of potentially un-
limited appeal to the cultural historian and may thus help liberate the study
of language from excessive isolation. A multi-level etymological essay,
elegantly phrased and saturated with literary implications (Baroque), such
as Jaberg's on the 'birthmark' (Romance Philology X [1956-1957], 307-
342), converts even the most militantly indifferent "humanist" to an
appreciation of linguistics.
In our curricula formal seminars on etymology, properly organized,
could raise a dormant interest to a high pitch of active curiosity. But only
at rare intervals do etymological cruxes qualify as thesis subjects, since they
present an uncomfortably high concentration of difficulties and presuppose
such balanced perspective as can be slowly gained from experience alone.
Will the refinement of etymological theory improve etymological
practice? Such a corpus of theory is already in existence, but in disap-
pointingly diffuse, inexplicit form.23 Rejection of theory as something
essentially sterile is not new (a mouthpiece of Goethe's advanced it in his
Faust, even more articulately in the Urfaust). Just as a keen esthetician
is not necessarily an overpowering artist, so an etymological theorist need
not be the most inspired elucidator of word origins. But at this critical
stage in the reorganization of human knowledge, etymology cannot be
redeemed without solid theoretical underpinning.
University of California at Berkeley

23 Cf. Etymologica: Festschrift fiir Walther von Wartburg (Ti.ibingen, 1958), passim.